CN110929596A - Shooting training system and method based on smart phone and artificial intelligence - Google Patents
Shooting training system and method based on smart phone and artificial intelligence Download PDFInfo
- Publication number
- CN110929596A CN110929596A CN201911080456.2A CN201911080456A CN110929596A CN 110929596 A CN110929596 A CN 110929596A CN 201911080456 A CN201911080456 A CN 201911080456A CN 110929596 A CN110929596 A CN 110929596A
- Authority
- CN
- China
- Prior art keywords
- detection
- basket
- player
- target
- basketball
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 116
- 238000012549 training Methods 0.000 title claims abstract description 47
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 20
- 238000001514 detection method Methods 0.000 claims abstract description 128
- 238000004458 analytical method Methods 0.000 claims abstract description 15
- 230000000007 visual effect Effects 0.000 claims abstract description 9
- 238000013135 deep learning Methods 0.000 claims description 40
- 230000009471 action Effects 0.000 claims description 21
- 238000013527 convolutional neural network Methods 0.000 claims description 11
- 238000000605 extraction Methods 0.000 claims description 11
- 238000013136 deep learning model Methods 0.000 claims description 8
- 230000008859 change Effects 0.000 claims description 7
- 210000002478 hand joint Anatomy 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 5
- 239000011159 matrix material Substances 0.000 claims description 5
- 230000003287 optical effect Effects 0.000 claims description 5
- 239000002245 particle Substances 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 5
- 238000002372 labelling Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- 125000004122 cyclic group Chemical group 0.000 claims description 3
- 238000012790 confirmation Methods 0.000 claims description 2
- 230000007547 defect Effects 0.000 abstract description 2
- 238000011897 real-time detection Methods 0.000 abstract 1
- 230000036544 posture Effects 0.000 description 9
- 238000010586 diagram Methods 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 230000001133 acceleration Effects 0.000 description 3
- 238000013145 classification model Methods 0.000 description 3
- 238000003708 edge detection Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 238000005452 bending Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 210000004247 hand Anatomy 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
- 210000000707 wrist Anatomy 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
- G06V40/28—Recognition of hand or arm movements, e.g. recognition of deaf sign language
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63B—APPARATUS FOR PHYSICAL TRAINING, GYMNASTICS, SWIMMING, CLIMBING, OR FENCING; BALL GAMES; TRAINING EQUIPMENT
- A63B69/00—Training appliances or apparatus for special sports
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
Landscapes
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Human Computer Interaction (AREA)
- Physical Education & Sports Medicine (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a shooting training system and method based on a smart phone and artificial intelligence. During training, the court positioning system carries out position detection on the boundary line, the middle line, the penalty line, the limit area and the penalty line of the court; the basket detection system is used for carrying out basket position detection and tracking and real-time detection based on target tracking; the player identity recognition system confirms the position and the identity of a player; the player motion analysis system acquires position information of body joint points of players; the basketball track recording system records the position of the ball and the size of the target; the goal detecting system judges whether a ball enters the basket or not; the visual playback system stores and plays back the obtained information and the frames of the video. The invention facilitates the shooting training of basketball trainers and improves the defects of the existing shooting training system.
Description
Technical Field
The invention relates to a shooting training system, in particular to a shooting training system and method based on a smart phone and artificial intelligence.
Background
The shooting training has a very large proportion in the basketball, but for the public, the shooting training mostly depends on a video tutorial to carry out simulation learning, and the feedback and the record of important data such as action standard or not, basketball parabola track record and hit rate record are lacked.
There are two main ways of current shooting training: firstly, a video device is used for collecting, manual recording is carried out during video playback or a one-to-one teaching mode is carried out through a line, but the mode is expensive, needs a large amount of manpower and is low in efficiency; secondly, adopt intelligent ball, court to add position sensor, utilize the mode record of sensor to shoot and the action record analysis of relevant training, this kind of mode need invest very big energy in the research and development of special sensor, and the price is higher, and the prevalence is very low, is unfavorable for vast basketball fan to use. Therefore, the problems of high price and inconvenient use are mainly faced at present. Therefore, it is very important to provide a basketball training system or method with intelligent record analysis, low cost and easy use for basketball trainers on a convenient device such as a smart phone.
Disclosure of Invention
The purpose of the invention is as follows: the invention provides a shooting training system and method based on a smart phone and artificial intelligence, aiming at the problems that intelligent training feedback of basketball is lacked, a large amount of artificial records are needed for basketball training data records, and the artificial records are inaccurate in the prior art.
The technical scheme is as follows: the invention relates to a shooting training system based on a smart phone and artificial intelligence, which comprises a court positioning system, a basket detection system, a player identity recognition system, a player action analysis system, a basketball track recording system, a goal detection system and a visual playback system;
the court positioning system is used for carrying out primary detection based on a contour extraction detection method, or adopting a convolutional neural network for detecting court sidelines to carry out position detection on the sidelines of the used court under a mobile phone image coordinate system; according to the detected relative positions of the person and the ball relative to the positions of the positioning sideline, the penalty line and the center line, the positions of the ball and the position of the person are positioned;
the basket detection system detects the position of the basket based on a target detection method or an outline extraction method, and tracks and detects the basket in real time based on a target tracking method;
the player identity recognition system preliminarily judges and establishes the player identity based on a face recognition method, and then confirms the position and the identity of a detection frame of a human body in an image coordinate system of the player by combining a human body target detection method and a target tracking method, so that the construction of a subsequent system is facilitated; wherein the detection frame position is the target position;
the player motion analysis system acquires the position of a player, the size of a target frame detected by the player and the identity of the player based on the player identity recognition system, and acquires the position information of body joint points of the player based on a posture estimation method according to the position information and the information of the size of the detection frame;
the basketball track recording system detects the basketball based on a target detection method and tracks and records the basketball track by using a target tracking method; recording the position of the ball and the size of the target based on the court positioning system;
the goal detecting system judges whether a ball enters the basket or not according to the court positioning and basket detecting system to obtain the position of the basket and the size of the detecting frame;
the visual playback system stores and plays back the obtained information and the frames of the video.
In the court positioning system, a convolutional neural network for detecting the court sidelines learns the sideline marking data of the boundary, the middle line, the penalty line, the limit area and the penalty area of the court through a deep learning model.
In the basket detection system, the target detection method is based on a deep learning method or a traditional target detection method; learning labeling data for labeling the basket for the deep learning model based on a deep learning method to obtain a basket detection model for basket detection; the traditional target detection method is basket detection by using a Harr characteristic + Adaboost + Cascade method or a HOG characteristic + SVM method.
The contour extraction method is characterized in that the shape and color characteristics of the basket are preset during detection, and similarity matching is performed on the shape and color characteristics in an image during use.
The target tracking method is a deep learning method or a traditional tracking method; the deep learning method is characterized in that a convolutional neural network in deep learning is utilized to learn the characteristics of a marked basketball target, and convolutional characteristic comparison similarity is extracted from the image of the current frame and the image of the previous frame to predict the position change of the target image; the traditional tracking method is based on a cyclic matrix and Fourier transform, and Meanshift, Particle Filter, Kalman Filter filtering, or an optical flow method based on characteristic points to predict and track the position of the basket target in the next frame image.
The human body target detection method is to use a deep learning method to learn the data labeled on the human body target to obtain a human body detection model for detecting the human body target.
The posture estimation method is to learn the labeled human key points by adopting a deep learning method.
The invention relates to a training method of a shooting training system based on a smart phone and artificial intelligence, which comprises the following steps:
(1) carrying out court positioning and basket detection, firstly erecting the smart phone, aligning a detection frame of a mobile phone screen to a basket, and automatically starting a basket detection method;
(2) carrying out player identity recognition, starting a face recognition method and a target detection and tracking method to carry out player identity confirmation and memory, and acquiring player positions and detection frame information;
(3) performing action analysis on the players, namely running a posture estimation method based on deep learning according to the positions of the players and the information of the detection frame to acquire two-dimensional or three-dimensional information of body joint points and hand joint points of the players; judging the action information of the player according to the distance and angle information of different joint points;
(4) recording the basketball track, and continuously detecting and tracking the basketball to acquire the position and the detection frame information of the basketball when the basketball appears in the screen;
(5) detecting the goal, namely judging whether a ball enters the basket or not according to the position of the basket and the size of a detection frame obtained by the court positioning and basket detection system;
(6) and (5) playing back the shooting training process, and storing and playing back the information and the frames of the video obtained in the steps (2) to (5).
And (3) operating a three-dimensional reconstruction method according to the court positioning information to acquire three-dimensional information of the human body joint points and the hand joint points.
Has the advantages that: compared with the existing basketball training system and method, the basketball training system has the following advantages:
(1) the shooting training system can meet the operation requirements of a computer and a cloud server, is subjected to operation optimization design and adaptation aiming at a system of a smart phone, such as android and IOS, and operation equipment, such as CPU, NPU and DSP, and can be applied to the smart phone, such as a mobile phone APP.
(2) The invention provides scientific training guidance and training data based on artificial intelligence, such as action analysis, scoring trajectory and shooting angle data recording analysis, which is convenient for basketball trainers to shoot, improves the defects of the existing shooting training system, and brings artificial intelligence into basketball training.
(3) The intelligent scientific guidance and the convenient and visual feedback are provided by depending on artificial intelligence and fusion of various algorithms, and the algorithm is transplanted into the smart phone and used at any time by utilizing an acceleration technology aiming at the smart phone, so that a brand-new training mode and experience are provided for a large number of basketball enthusiasts.
Drawings
FIG. 1 is a system block diagram of the present invention;
FIG. 2 is an example illustration of the present invention;
FIG. 3 is an explanatory diagram of the manner of installing the smart phone of the present invention;
FIG. 4 is a schematic diagram of the use of the ball frame detection in the present invention;
FIG. 5 is a diagram of a gesture joint in the present invention;
FIG. 6 is a schematic view of a human joint point of the present invention;
FIG. 7 is a schematic view of the shot-to-hand angle and speed of the present invention;
fig. 8 is a schematic view of the shooting position and hit condition visualization in the invention.
Detailed Description
As shown in fig. 1, the shooting training system of the invention comprises a court positioning system, a basket detection system, a player identification system, a player action analysis system, a basketball track recording system, a goal detection system and a visual playback system; the order and relationship between the various systems of the above system may be adjusted as desired.
The court positioning system is constructed based on a traditional contour extraction detection method to carry out preliminary detection, or a convolutional neural network for detecting the boundary line, the middle line, the penalty line, the limit area and the penalty area of a court is adopted to carry out position detection under a mobile phone image coordinate system, and a user can drag the boundary line on a mobile phone to position according to the detection effect, so that the positions of the boundary line, the middle line and the penalty line of the court under a pixel coordinate system become known, and the positions of a ball and a person can be positioned according to the relative positions of the detected position of the person and the position of the ball relative to the positioning boundary line, the penalty line and the middle line.
And the convolutional neural network for edge detection learns the edge marking data of the boundary, the middle line, the penalty line, the limit zone and the penalty zone of the court through the deep learning model so as to obtain an edge detection model for court edge detection.
The basket detection system detects the position of the basket based on a target detection method or an outline extraction method, and tracks and detects the basket in real time based on a target tracking method. The target detection method is based on a deep learning method, namely, a basket detection model is obtained by learning the labeling data of the labeled basket by adopting a deep learning model for basket detection.
The traditional method is to use Harr characteristic + Adaboost + Cascade method or HOG characteristic + SVM method to carry out basket detection.
When the outline extraction method is used for carrying out the position detection of the basket, namely the shape and the color of the basket are preset during the detection, and when the outline extraction method is used, the similarity matching is carried out on the shape and the color of the basket in the image, so that the position detection of the basket is further carried out.
The target tracking method is a deep learning method or a traditional tracking method, wherein the deep learning method is to learn the characteristics of the marked basketball target by using a convolutional neural network in deep learning, and extract the convolutional characteristic comparison similarity of the image of the current frame and the image of the previous frame when in use to predict the position change of the target image. The traditional tracking method is realized based on a circulant matrix and Fourier transform, and a filtering method of Meanshift, Particle Filter and Kalman Filter, or an optical flow method based on characteristic points is used for predicting the position of the basket target in the next frame image, so that the basket is tracked and detected in real time.
The player identity recognition system preliminarily judges and establishes the player identity based on a face recognition method, and then confirms the position and the identity of a detection frame of a human body in an image coordinate system of the player by combining a human body target detection method and a target tracking method, so that the construction of a subsequent system is facilitated; where the frame position, i.e. the target position, is detected.
The human body target detection method is to use a deep learning method to learn the data labeled on the human body target to obtain a human body detection model for detecting the human body target.
The target tracking method comprises a deep learning method and a traditional method, wherein the deep learning method is to learn the characteristics of the annotated basketball target by using a convolutional neural network in the deep learning, and extract the convolutional characteristic comparison similarity of the image of the current frame and the image of the previous frame when in use so as to predict the position change of the target image. The traditional method is a method based on a cyclic matrix and Fourier transform, a filtering method of Meanshift, Particle Filter and Kalman Filter, or an optical flow method based on characteristic points to predict and track the position of a human target in the next frame of image.
The player motion analysis system is constructed based on the position of a player, the size of a target frame detected by the player and the identity of the player which are obtained by the player identity recognition system, and the position information of body joint points of the player is obtained based on a posture estimation method according to the position information and the information of the size of the detection frame. Meanwhile, the position and size information of a detection frame of the basketball can be obtained by using a basketball detection technology, different actions of the player are analyzed and judged according to the position information of the joint point and the position and size information of the basketball detection frame, so that all action information related to the body of the player, such as the shooting angle, the shooting point height, the angles and postures of all parts of the body at different moments and positions, the height of feet from the ground and the like, of the player can be obtained, specifically, the action track of the player can be restored by recording and analyzing key points of the human body, for example, when the player lifts the left arm, the key points of the elbow and the wrist have certain motion tracks, the action is judged and analyzed according to the motion tracks, and more actions can be analyzed by combining the detection frame information of the basketball; the action analysis system of the player is constructed and completed.
The posture estimation method is used for predicting key points of a human body by learning labeled key points of the human body by adopting a deep learning method to obtain a model for predicting the key points of the human body.
The basketball track recording system detects the basketball based on a target detection method and tracks and records the basketball track by using a target tracking method; and the position and the target size of the ball are recorded based on the court positioning system, and the motion trail of the ball can be recorded according to the method.
The target detection method is a deep learning method or a traditional target detection method; the deep learning method is used for learning and marking the marking data of the basketball by adopting a deep learning model to obtain a basketball detection model for basket detection. The traditional target detection method is to detect the basketball by using a Harr characteristic + Adaboost + Cascade method or a HOG characteristic + SVM method.
The target tracking method is a deep learning method or a traditional target tracking method; the deep learning method is characterized in that a convolutional neural network in deep learning is utilized to learn the features of the labeled basketball target, and convolutional feature comparison similarity is extracted from the image of the current frame and the image of the previous frame when the deep learning method is used, so that the position change of the target image is predicted. The traditional target tracking method is realized based on a circulant matrix and Fourier transform, and a filtering method of Meanshift, Particle Filter and Kalman Filter, or an optical flow method based on characteristic points is used for predicting and tracking the position of a basketball target in a next frame image, and tracking and recording the basketball track. The basketball track is formed by combining continuous position information of the basketball detection frame in the video.
The goal detection system obtains the position and the size of the basket according to the court positioning system and the basket detection system, compares and judges the position and the size of the basket with the target and the size of the basket according to the obtained position and the target size of the basketball, and can judge that the goal is achieved if the target basket of the basketball is completely framed by the target basket of the basket, or judge whether the goal is achieved by using a goal classification deep learning model for scenes in which the basketball detection frame is completely framed by the basket detection frame according to a deep learning method, so that whether the goal is achieved can be judged.
The visual playback system obtains the information of the goal time according to the relevant information of the goal in a section of video obtained by the goal detection system, obtains the shooting and hand-out time information of a player and the time information of the shooting action according to the player action analysis system, and captures and stores the relevant time in a section of video according to the information, if the section of video from the shooting and hand-out to the basketball goal frame detection time is stored as the shooting and hand-out playback video, the basketball track information obtained by the basketball track recording system is drawn in the video, meanwhile, the position of the player in the court at the hand-out time is obtained according to the court positioning system, and the hit rate recording diagrams and hit rate display of the basketball hand-out at different positions are drawn according to the result of whether the goal detection system enters the basketball.
The shooting training method based on the smart phone and the artificial intelligence comprises the following steps:
(1) carrying out court positioning and basket detection;
(2) carrying out player identity identification;
(3) analyzing the action of the player;
(4) recording the basketball track;
(5) detecting goal;
(6) and playing back the shooting training process.
As shown in FIG. 2, the shooting training process of the present invention is as follows:
in step 1, before court location and basket detection, the smart phone needs to be erected, as shown in fig. 3, for convenient use, three erection modes are provided for the smart phone:
(a) a non-training player a holds a smart phone b to align with a court and a ball frame to be used;
(b) erecting a mobile phone by using a mobile phone support c, and aligning a court and a ball frame to be used;
(c) the smart phone is started to have a front camera shooting function, and objects which are placed on the ground and cannot fall down are used for keeping the lens aligned with a court and a ball frame to be used.
After the smart phone is erected, the court positioning and basket detection are carried out, as shown in fig. 4, a basket detection method is automatically started by keeping a detection frame consisting of 8 small rectangles in a screen of the smart phone aligned with a basket to be used or ensuring that the basket to be used appears in the screen, and the detection method adopts a deep learning method or a traditional target detection method.
At the moment, if the smart phone is erected in a handheld and supporting mode, a boundary line, a middle line and a penalty line sideline detection method is started through a deep learning method, a traditional detection method or a contour extraction method; if the detection effect is not accurate, the court positioning can be completed by manually adjusting; if the smart phone is erected in the step 1 in a mode of leaning against other objects on the ground, a player is required to shoot a basketball at the position of the penalty line once, and the shot position of the player is regarded as the position of the penalty line, so that the course positioning system is roughly constructed.
And 2, identifying the identity of the player, starting a face identification method and a target detection and tracking method to confirm and memorize the identity of the player after the player walks into the mobile phone shooting range, and simultaneously detecting the position of the player and the size of the detection frame in real time, so far, as long as the position of the player at any moment in the mobile phone shooting range and the size of the detection frame are in a known state in the training. By the mode, a plurality of people can shoot simultaneously.
And 3, analyzing the movement of the player, acquiring the position of the player and the information of the detection frame according to the step 2, operating a posture estimation method based on deep learning according to the information, and acquiring 2D (two-dimensional) information of body joint points and hand joint points of the player, wherein the 2D information is X-axis direction information and Y-axis direction information only in an image coordinate system.
As shown in fig. 5 and 6, the three-dimensional reconstruction method can be operated based on the course positioning information in step 1 to obtain 3D information of the human body joint points and the hand joint points, where the 3D information is information in the X-axis direction, information in the Y-axis direction, and information in the Z-axis direction in the real world coordinate system, specifically, points with serial numbers of 0 to 20 in fig. 5 are defined as preset points for training the gesture key point detection model, and points with serial numbers of 0 to 20 of the hand players are predicted through the gesture key point model during subsequent use. And judging the action information of the player at the moment according to the distance and angle information of different joint points. Specifically, the distance between the ball and the human body is judged by combining the player detection frame information and the basketball detection frame information in the steps 2 and 4 when judging whether the player shoots and shoots the hand and the player goes to the basketball, and the time when the player shoots or not or the time when the player shoots the hand is in the process of shooting is comprehensively judged by combining the hand-shooting motion classification model algorithm based on deep learning or machine learning. After the relevant conditions of the shooting and the shooting of the basketball are determined, the change trend of the joint points of the gesture and the relation between the joint points are further analyzed to obtain shooting gestures, shooting angles and initial speed information of the basketball. As shown in fig. 7: s represents the horizontal distance of the starting point from the basket, theta represents the basketball starting angle,
representing the basketball hand-out speed, and h representing the vertical height of the hand-out point from the basket; the information such as the bending angle of the legs, the body posture, the height above the ground when the hands are lifted and the like is judged according to the joint points of the legs and the trunk of the human body.
And 4, recording the basketball track, when the basketball appears in the screen, continuously tracking and detecting the basketball and giving the basketball a tracking 'identity information' by using a target detection method, namely a deep learning method or a traditional method, and a target tracking method, namely a traditional method or a deep learning method, in the system, and automatically ignoring other new basketballs after entering a camera shooting range because only the first detected basketball is tracked. And (3) continuously detecting and tracking the basketball to acquire the position of the basketball and the 2D information of the detection frame, positioning the court in the step 1 to perform three-dimensional reconstruction, and acquiring the 3D information of the position of the basketball and the detection frame.
And 5, detecting the goal, and when the basketball detection frame information and the detection frame information of the frame meet the conditions, operating a goal detection algorithm, such as a method adopting geometric area calculation or a goal classification model method based on deep learning, specifically, obtaining a goal classification model by utilizing a picture of the basket and the basketball marked as the goal moment in the deep learning training so as to judge whether the goal is achieved, and performing goal detection so as to obtain the category information whether the goal is achieved by the basketball at a certain time.
And 6, storing the information obtained in the steps 1, 2, 3, 4 and 5 and the frames of the related videos by a visual playback system, wherein the information can be specifically stored as shooting posture playback, shooting position and shooting goal recording picture playback. As shown in fig. 8, a solid origin indicates a goal of shooting at this position, and an open cross point indicates no goal of the player at this position. Similarly, the time from the hand-out to the goal or no goal for each shot is recorded and can be played back.
The system and the method can run on the smart phone because of using the acceleration technology aiming at the smart phone, and can be used by an offline host or a cloud connected with a camera if the acceleration technology aiming at the smart phone is not adopted, so that the use scene of the system and the change of the use equipment are both in the scope of the right of the invention; in addition, the above example is only directed to one application sub-scenario illustrating the functionality of the system, and other application scenarios and situations derived therefrom, as well as any functionality of the system, are within the scope of the present invention.
Claims (9)
1. The utility model provides a shooting training system based on smart mobile phone and artificial intelligence which characterized in that: the basketball goal tracking system comprises a court positioning system, a basket detection system, a player identity recognition system, a player action analysis system, a basketball track recording system, a goal detection system and a visual playback system;
the court positioning system is used for carrying out primary detection based on a contour extraction detection method, or adopting a convolutional neural network for detecting court sidelines to carry out position detection on the sidelines of the used court under a mobile phone image coordinate system; according to the detected relative positions of the person and the ball relative to the positioned sidelines, the penalty lines and the center lines, the positions of the ball and the positions of the person are positioned;
the basket detection system is used for detecting the position of the basket based on a target detection method or an outline extraction method, and tracking and detecting the basket in real time based on a target tracking method;
the player identity recognition system preliminarily judges and establishes the player identity based on a face recognition method, and then confirms the position and the identity of a detection frame of a human body in an image coordinate system of the player by combining a human body target detection method and a target tracking method, so that the construction of a subsequent system is facilitated; the position of the detection frame is the position of the target under the image coordinate system;
the player motion analysis system acquires the position of a player, the size of a target frame detected by the player and the identity of the player based on the player identity recognition system, and acquires the position information of body joint points of the player based on a posture estimation method according to the position information and the information of the size of the detection frame;
the basketball track recording system detects a basketball based on a target detection method and tracks and records the basketball track by using a target tracking method; recording the position of the ball and the size of the target based on the court positioning system;
the goal detecting system judges whether a ball enters the basket or not according to the court positioning and basket detecting system to obtain the position of the basket and the size of the detecting frame;
and the visual playback system stores and plays back the obtained information of the goal moment, the shooting and hand-out moment of the player, the moment information of the action on the basket and the frame of the video.
2. The smartphone and artificial intelligence based shooting training system of claim 1, wherein: in the court positioning system, a convolutional neural network for detecting the court sidelines learns the sideline marking data of the boundary, the middle line, the penalty line, the limit area and the penalty area of the court through a deep learning model.
3. The smartphone and artificial intelligence based shooting training system of claim 1, wherein: in the basket detection system, the target detection method is based on a deep learning method or a traditional target detection method; the basket detection model is used for basket detection and is obtained by learning the labeling data of the labeled basket for the deep learning model based on the deep learning method; the traditional target detection method is basket detection by using a Harr characteristic + Adaboost + Cascade method or a HOG characteristic + SVM method.
4. The smartphone and artificial intelligence based shooting training system of claim 1, wherein: the contour extraction method is characterized in that the shape and color characteristics of the basket are preset during detection, and similarity matching is performed on the shape and color characteristics in an image during use.
5. The smartphone and artificial intelligence based shooting training system of claim 1, wherein: the target tracking method is a deep learning method or a traditional tracking method; the deep learning method is characterized in that a convolutional neural network in deep learning is utilized to learn the characteristics of a marked basketball target, and convolutional characteristic comparison similarity is extracted from the image of the current frame and the image of the previous frame to predict the position change of the target image; the traditional tracking method is based on a cyclic matrix and Fourier transform, and Meanshift, Particle Filter, Kalman Filter filtering, or an optical flow method based on characteristic points to predict and track the position of the basket target in the next frame image.
6. The smartphone and artificial intelligence based shooting training system of claim 1, wherein: the human body target detection method is used for learning the data marked on the human body target by utilizing a deep learning method to obtain a human body detection model for detecting the human body target.
7. The smartphone and artificial intelligence based shooting training system of claim 1, wherein: the posture estimation method is used for learning the labeled human key points by adopting a deep learning method.
8. A training method using the smartphone-based and artificial intelligence-based shooting training system of any one of claims 1-7, characterized by: the method comprises the following steps:
(1) carrying out court positioning and basket detection, firstly erecting the smart phone, aligning a detection frame of a mobile phone screen to a basket, and automatically starting a basket detection method;
(2) carrying out player identity recognition, starting a face recognition method and a target detection and tracking method to carry out player identity confirmation and memory, and acquiring player positions and detection frame information;
(3) performing action analysis on the players, namely running a posture estimation method based on deep learning according to the positions of the players and the information of the detection frame to acquire two-dimensional or three-dimensional information of body joint points and hand joint points of the players; judging the action information of the player according to the distance and angle information of different joint points;
(4) recording the basketball track, and continuously detecting and tracking the basketball to acquire the position and the detection frame information of the basketball when the basketball appears in the screen;
(5) detecting the goal, namely judging whether a ball enters the basket or not according to the position of the basket and the size of a detection frame obtained by the court positioning and basket detection system;
(6) and (5) playing back the shooting training process, and storing and playing back the information and the frames of the video obtained in the steps (2) to (5).
9. The smartphone and artificial intelligence based shooting training method of claim 8, wherein: and (3) operating a three-dimensional reconstruction method according to the court positioning information to acquire three-dimensional information of the human body joint points and the hand joint points.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911080456.2A CN110929596A (en) | 2019-11-07 | 2019-11-07 | Shooting training system and method based on smart phone and artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911080456.2A CN110929596A (en) | 2019-11-07 | 2019-11-07 | Shooting training system and method based on smart phone and artificial intelligence |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110929596A true CN110929596A (en) | 2020-03-27 |
Family
ID=69853640
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911080456.2A Pending CN110929596A (en) | 2019-11-07 | 2019-11-07 | Shooting training system and method based on smart phone and artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110929596A (en) |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111368810A (en) * | 2020-05-26 | 2020-07-03 | 西南交通大学 | Sit-up detection system and method based on human body and skeleton key point identification |
| CN111444890A (en) * | 2020-04-30 | 2020-07-24 | 汕头市同行网络科技有限公司 | Sports data analysis system and method based on machine learning |
| CN111461020A (en) * | 2020-04-01 | 2020-07-28 | 浙江大华技术股份有限公司 | Method and device for identifying behaviors of insecure mobile phone and related storage medium |
| CN111476291A (en) * | 2020-04-03 | 2020-07-31 | 南京星火技术有限公司 | Data processing method, device and storage medium |
| CN111626155A (en) * | 2020-05-14 | 2020-09-04 | 新华智云科技有限公司 | Basketball position point generation method and equipment |
| CN111680608A (en) * | 2020-06-03 | 2020-09-18 | 长春博立电子科技有限公司 | Intelligent sports auxiliary training system and training method based on video analysis |
| CN111724414A (en) * | 2020-06-23 | 2020-09-29 | 宁夏大学 | Basketball movement analysis method based on 3D attitude estimation |
| CN112802051A (en) * | 2021-02-02 | 2021-05-14 | 新华智云科技有限公司 | Fitting method and system of basketball shooting curve based on neural network |
| CN113033384A (en) * | 2021-03-23 | 2021-06-25 | 清华大学 | Wheelchair curling motion state detection and target tracking system |
| CN113058246A (en) * | 2021-03-23 | 2021-07-02 | 清华大学 | Wheelchair curling track identification, positioning, tracking and motion state detection system |
| CN113181613A (en) * | 2021-01-25 | 2021-07-30 | 河南质量工程职业学院 | Sports teaching is with dribbling shooting training system |
| CN113449822A (en) * | 2021-08-31 | 2021-09-28 | 长沙鹏阳信息技术有限公司 | Configurable GOA system for rapid artificial intelligence application development |
| CN113537168A (en) * | 2021-09-16 | 2021-10-22 | 中科人工智能创新技术研究院(青岛)有限公司 | Basketball goal detection method and system for rebroadcasting and court monitoring scene |
| CN114980984A (en) * | 2020-09-08 | 2022-08-30 | 丹尼斯·阿多马科 | Basketball shooting device |
| CN114994046A (en) * | 2022-04-19 | 2022-09-02 | 深圳格芯集成电路装备有限公司 | Defect detection system based on deep learning model |
| CN116109981A (en) * | 2023-01-31 | 2023-05-12 | 北京智芯微电子科技有限公司 | Shooting recognition method, basketball recognition device, electronic equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105664462A (en) * | 2016-01-07 | 2016-06-15 | 北京邮电大学 | Auxiliary training system based on human body posture estimation algorithm |
| CN107303428A (en) * | 2016-04-20 | 2017-10-31 | 李斌 | Basketball goal decision method and system based on image procossing |
| CN109766768A (en) * | 2018-12-18 | 2019-05-17 | 南京邮电大学 | A kind of sportsman's recognition methods based on deep learning |
| CN110298231A (en) * | 2019-05-10 | 2019-10-01 | 新华智云科技有限公司 | A kind of method and system determined for the goal of Basketball Match video |
-
2019
- 2019-11-07 CN CN201911080456.2A patent/CN110929596A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105664462A (en) * | 2016-01-07 | 2016-06-15 | 北京邮电大学 | Auxiliary training system based on human body posture estimation algorithm |
| CN107303428A (en) * | 2016-04-20 | 2017-10-31 | 李斌 | Basketball goal decision method and system based on image procossing |
| CN109766768A (en) * | 2018-12-18 | 2019-05-17 | 南京邮电大学 | A kind of sportsman's recognition methods based on deep learning |
| CN110298231A (en) * | 2019-05-10 | 2019-10-01 | 新华智云科技有限公司 | A kind of method and system determined for the goal of Basketball Match video |
Non-Patent Citations (1)
| Title |
|---|
| 谢剑斌, 国防科技大学出版社 * |
Cited By (25)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111461020A (en) * | 2020-04-01 | 2020-07-28 | 浙江大华技术股份有限公司 | Method and device for identifying behaviors of insecure mobile phone and related storage medium |
| CN111461020B (en) * | 2020-04-01 | 2024-01-19 | 浙江大华技术股份有限公司 | Recognition method, equipment and related storage medium for unsafe mobile phone behavior |
| CN111476291B (en) * | 2020-04-03 | 2023-07-25 | 南京星火技术有限公司 | Data processing method, device and storage medium |
| CN111476291A (en) * | 2020-04-03 | 2020-07-31 | 南京星火技术有限公司 | Data processing method, device and storage medium |
| CN111444890A (en) * | 2020-04-30 | 2020-07-24 | 汕头市同行网络科技有限公司 | Sports data analysis system and method based on machine learning |
| CN111626155B (en) * | 2020-05-14 | 2023-08-01 | 新华智云科技有限公司 | Basketball position point generation method and equipment |
| CN111626155A (en) * | 2020-05-14 | 2020-09-04 | 新华智云科技有限公司 | Basketball position point generation method and equipment |
| CN111368810A (en) * | 2020-05-26 | 2020-07-03 | 西南交通大学 | Sit-up detection system and method based on human body and skeleton key point identification |
| CN111368810B (en) * | 2020-05-26 | 2020-08-25 | 西南交通大学 | Sit-up detection system and method based on human body and skeleton key point identification |
| CN111680608A (en) * | 2020-06-03 | 2020-09-18 | 长春博立电子科技有限公司 | Intelligent sports auxiliary training system and training method based on video analysis |
| CN111680608B (en) * | 2020-06-03 | 2023-08-18 | 长春博立电子科技有限公司 | Intelligent sports auxiliary training system and training method based on video analysis |
| CN111724414A (en) * | 2020-06-23 | 2020-09-29 | 宁夏大学 | Basketball movement analysis method based on 3D attitude estimation |
| CN111724414B (en) * | 2020-06-23 | 2024-01-26 | 宁夏大学 | Basketball motion analysis method based on 3D gesture estimation |
| CN114980984A (en) * | 2020-09-08 | 2022-08-30 | 丹尼斯·阿多马科 | Basketball shooting device |
| CN113181613A (en) * | 2021-01-25 | 2021-07-30 | 河南质量工程职业学院 | Sports teaching is with dribbling shooting training system |
| CN112802051A (en) * | 2021-02-02 | 2021-05-14 | 新华智云科技有限公司 | Fitting method and system of basketball shooting curve based on neural network |
| CN112802051B (en) * | 2021-02-02 | 2022-05-17 | 新华智云科技有限公司 | Fitting method and system of basketball shooting curve based on neural network |
| CN113033384A (en) * | 2021-03-23 | 2021-06-25 | 清华大学 | Wheelchair curling motion state detection and target tracking system |
| CN113058246A (en) * | 2021-03-23 | 2021-07-02 | 清华大学 | Wheelchair curling track identification, positioning, tracking and motion state detection system |
| CN113449822A (en) * | 2021-08-31 | 2021-09-28 | 长沙鹏阳信息技术有限公司 | Configurable GOA system for rapid artificial intelligence application development |
| CN113537168B (en) * | 2021-09-16 | 2022-01-18 | 中科人工智能创新技术研究院(青岛)有限公司 | Basketball goal detection method and system for rebroadcasting and court monitoring scene |
| CN113537168A (en) * | 2021-09-16 | 2021-10-22 | 中科人工智能创新技术研究院(青岛)有限公司 | Basketball goal detection method and system for rebroadcasting and court monitoring scene |
| CN114994046A (en) * | 2022-04-19 | 2022-09-02 | 深圳格芯集成电路装备有限公司 | Defect detection system based on deep learning model |
| CN116109981A (en) * | 2023-01-31 | 2023-05-12 | 北京智芯微电子科技有限公司 | Shooting recognition method, basketball recognition device, electronic equipment and storage medium |
| CN116109981B (en) * | 2023-01-31 | 2024-04-12 | 北京智芯微电子科技有限公司 | Shooting recognition method, basketball recognition device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110929596A (en) | Shooting training system and method based on smart phone and artificial intelligence | |
| US11638854B2 (en) | Methods and systems for generating sports analytics with a mobile device | |
| US11842572B2 (en) | Device, system, and method of computer vision, object tracking, image analysis, and trajectory estimation | |
| CN108564596B (en) | Intelligent comparison analysis system and method for golf swing video | |
| CN107281709B (en) | A kind of extracting method and device, electronic equipment of sport video segment | |
| CN112183355B (en) | Effluent height detection system and method based on binocular vision and deep learning | |
| EP2707837A1 (en) | Method of analysing a video of sports motion | |
| CN105426827A (en) | Living body verification method, device and system | |
| Teachabarikiti et al. | Players tracking and ball detection for an automatic tennis video annotation | |
| CN101826155B (en) | Method for identifying act of shooting based on Haar characteristic and dynamic time sequence matching | |
| CN109684919B (en) | Badminton service violation distinguishing method based on machine vision | |
| CN115624735B (en) | Auxiliary training system for ball games and working method | |
| CN106991358A (en) | The algorithm that automatic identification football match based on panoramic video is scored | |
| Sha et al. | Understanding and analyzing a large collection of archived swimming videos | |
| US10025986B1 (en) | Method and apparatus for automatically detecting and replaying notable moments of a performance | |
| Sokolova et al. | Human identification by gait from event-based camera | |
| CN114140721A (en) | Archery posture evaluation method and device, edge calculation server and storage medium | |
| CN110929595A (en) | System and method for training or entertainment with or without ball based on artificial intelligence | |
| CN116758459A (en) | Real-time analysis system and method for badminton match | |
| CN113837063A (en) | Curling motion field analysis and decision-making assisting method based on reinforcement learning | |
| CN114639117A (en) | Cross-border specific pedestrian tracking method and device | |
| CN113673327A (en) | Penalty ball hit prediction method based on human body posture estimation | |
| CN111265835A (en) | Throwing exercise training equipment and control method thereof | |
| CN110674769A (en) | Deep learning algorithm-based classification method for non-goal, common goal and location ball of football video clip | |
| Akan et al. | Reidentifying soccer players in broadcast videos using Body Feature Alignment Based on Pose |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200327 |